Compartmental signal modulation: Endosomal phosphatidylinositol 3-phosphate controls endosome morphology and selective cargo sorting

Abstract

It is increasingly recognized that the compartmental organization of signaling processes has a profound influence on cellular behavior. However, our inability to influence these compartmental events in a spatially restricted and acute manner limits our understanding of causation. To determine whether local compartmental loss of a phosphoinositide disrupts the normal traffic of specific cargoes through endosomes, we developed the use of a regulated dimerization device, here designed to compartmentally modify the phosphoinositide content of Rab5-positive endosomes. This modification is effected through the specific regulated recruitment of the 3-phosphatase myotubularin to endosomal membranes in intact cells. The selective manipulation of endosomal phosphatidylinositols (PIs) demonstrates that it is the phosphatidylinositol 3-phosphate (PtdIns3P) or its metabolite PtdIns(3,5)P2 within this compartment that determines the normal maturation of the endosomal compartment and the flux of receptors through it. On local loss of PtdIns3P/PtdIns(3,5)P2, the endosomal compartment itself fails to continue its normal maturation process, leading to the microtubule-dependent tubularization of the endosomal network. Furthermore, it is shown that endosomal PtdIns3P/PtdIns(3,5)P2 is necessary for transferrin receptor traffic through this compartment while having an effect on EGF receptor (EGFR) entry into and sorting from this endosome compartment. The ability to acutely and selectively influence compartmental behavior as exemplified here for endomsomes clearly illustrates the power of the approach used to dissect the role of localized signals and events.

Fig. 1.

In vitro and in vivo validation of the heterodimerization device. (a) Principle of the dimerization system. Based on Ariad's heterodimerization kit, the system exploits the natural rapamycin-induced dimerization of FKBP with a mutated portion of mTOR/FRAP (FRB*). To target the endosomal pool of PtdIns3P, the early endosomal marker hRab5a and the PtdIns3P-specific 3-phosphatase myotubularin (hMTM1wt) have been fused to 2xFKBP and FRB, respectively. The nonimmunosuppressive analogue of rapamycin (rapalogue) was used to induce heterodimerization and therefore recruitment of myotubularin to the Rab5a compartment. (b) Time course of FKBP-FRB complex formation. HeLa cells were transiently cotransfected with EGFP-2xFKBP-hRab5a (GFP-Rab5a-target) and mRFP-Flag-FRB-hMTM1wt (RFP-MTM-recruit) for 24 h, treated with or without 500 nM rapalogue for the indicated time, and lysed with Triton X-100. GFP-Rab5a-target was immunoprecipitated by using anti-GFP antibody. Coimmunoprecipitated RFP-MTM-recruit was detected by anti-Flag antibody. (c and d) The immunoprecipitated FKBP-FRB* complex specifically dephosphorylates PtdIns3P and PtdIns(3,5)P₂ in a rapalogue-dependent manner. Cos7 cells transiently coexpressing GFP-Rab5a-target and RFP-MTM-recruit were treated for 20 min with or without rapalogue and lysed with Triton X-100. The FKBP/FRB* complex was immunoprecipitated by using anti-GFP antibody, and its 3-phosphatase activity was assessed by using 100 μM different C16-PI substrates during a 30-min reaction at 37°C. Released free phosphate was detected by using the malachite green assay (see Supporting Materials and Methods, which is published as supporting information on the PNAS web site) (n = 3 experiments). (e) The 3-phosphatase activity of the FKBP/FRB complex depends on the presence and activity hMTM1wt. The 3-phosphatase activity of the WT complex was confirmed by the inhibitory effect of the vanadate–derived 3-phosphatase inhibitor bpV (OHpic) (100 μM). The assay was repeated for the FKBP/FRB complex containing the mRFP-Flag-FRB (RFP-FRB) mutant lacking hMTM1wt and the phosphatase inactive mRFP-Flag-FRB-MTM1C375S (RFP-MTM*-recruit) mutant (n = 3 experiments). Coimmunoprecipitation of all of the FRB-containing proteins was detected by anti-flag antibody. (f) MTM1 gets specifically recruited on the Rab5a-positive compartment in a rapalogue-dependent manner. HeLa cells were transiently transfected with GFP-Rab5a-target (green) and RFP-MTM-recruit (red) for 24 h and treated for 5 min with or without 500 nM rapalogue. The colocalization of Rab5a and MTM1 is shown in yellow. Confocal z-stack projections of representative cells are shown (n = 3 experiments). (Scale bar: 10 μm.)

Fig. 2.

Tubularization is due to the localized decrease of the endosomal PtdIns3P. A recombinant GST-2xFYVE_Hrs PtdIns3P-specific probe was used to detect the status of endogenous PtdIns3P. HeLa cells expressing GFP-Rab5a-target (green) and RFP-MTM-recruit (white) (a) or RFP-FRB (b) were treated for 90 min with or without the rapalogue, fixed with 4% paraformaldehyde, and permeabilized with 50 μg/ml digitonin. Localization of the GST-2xFYVE_Hrs probe was assessed by indirect immunofluorescence confocal microscopy by using mouse anti-GST antibody combined with Cy5-conjugated (red) anti-mouse antibody. The asterisk indicates where the GST-2xFYVE_Hrs staining is significantly decreased as a result of the rapalogue-induced tubularization of the Rab5a-positive compartment. Confocal z-stack projections of representative cells are shown. (Scale bar: 10 μm.)

Fig. 3.

hMTM1_wt recruitment induces tubularization of the Rab5a compartment in a microtubule-dependent manner. (a) HeLa cells expressing GFP-Rab5a-target (green) and RFP-MTM-recruit were stimulated with rapalogue and imaged live. Frames were captured every 23 s for 1 h 23 min. Frames from the Inset correspond to 30 min after rapalogue stimulation and depict the formation of the tubularized Rab5a-positive endosomes. The numbers indicate regions where tubularization events occur. (b) HeLa cells expressing GFP-Rab5a-target and RFP-MTM-recruit were treated with rapalogue as indicated. To assess the occurrence of tubularization over time, the percentage of cells (out of 50 cells per condition) was calculated and normalized to the maximum percent tubularization. The average value from the indicated number of experiments is represented. (c and d) HeLa cells expressing GFP-Rab5a-target (green) and RFP-MTM-recruit (red) were pretreated for 30 min with the microtubule-depolymerizing drug nocodazole (20 μM) and incubated for 90 min with or without rapalogue. (c) The microtubule network (blue) was detected by indirect immunofluorescence by using Cy5-conjugated anti-mouse antibody combined with anti-tubulin antibody. Confocal z-stack projections of representative cells are presented. (Scale bar: 10 μm.) (d) The inhibitory effect of nocodazole on tubularization was quantified and represented as the percentage of cells containing tubularized structures (n = 3, 150 cells per experiment). The endosomal compartment was considered to be tubularized when there was a minimum of two tubules of ≈50 μm.

Fig. 4.

Tubularized endosomes displace SNX1 and prevent recycling of transferrin. (a–c) HeLa cells expressing GFP-Rab5a-target (green) and RFP-MTM-recruit (white) were treated for 90 min with or without rapalogue, fixed, and stained with antibodies against EEA1 (a), SNX1 (b), or TfnR (c) combined with the Cy5-conjugated (red) secondary antibody. The asterisk in the SNX1 panel indicates a cell containing the rapalogue-induced tubularized endosomes. EEA1 and TfnR strongly stain the structures; SNX1 is completely displaced. Confocal z-stack projections of representative cells are shown. (Scale bar: 10 μm.) (d and e) Early endosomes in transfected HeLa cells were loaded with Tfn or EGF by incubating for 5 min with 30 μg/ml Tfn-Alexa Fluor 647 (red) or 10 min with 1 μg/ml EGF-Alexa Fluor 647 (red) at 37°C, respectively. Excess ligand was washed off, and dimerization was induced at 15°C as described in Materials and Methods. Tfn or EGF were chased at 37°C for different times. (d) After a 30-min chase, we compared the localization of endocytosed Tfn- and EGF-Alexa Fluor 647. Although EGF was retained on the main body of tubularized endosomes, Tfn was found on both the tubular and vesicular regions of the network. (e) To assess the effect of tubularization of Tfn traffic, Tfn-Alexa Fluor 647 was chased for longer time periods. The intensity of the Tfn signal was assessed by using the same parameters for image acquisition. Arrowheads indicate nontransfected cells, and the asterisk identifies a cell expressing only GFP-Rab5a-target. (Scale bar: 10 μm.)

Fig. 5.

Endosome tubularization delays but does not block EGF traffic. (a) HeLa cells coexpressing GFP-Rab5a-target (green) and RFP-MTM-recruit were treated as above with the rapalogue and incubated for 2.5 min with 1 μg/ml EGF-Alexa Fluor 647 (red). Unbound and surface bound ligand was washed off with ice-cold low-pH buffer (see Materials and Methods). The labeled EGF was chased at 37°C as indicated. Localization of the ligand was assessed by triple channel confocal fluorescence microscopy. (b) To label early endosomes with EGF, HeLa cells coexpressing GFP-Rab5a-target (green) and RFP-MTM-recruit (white) were incubated for 10 min at 37°C with 1 μg/ml EGF-Alexa Fluor 647 (red). Tubularization was induced as in Fig. 4e. EGF-Alexa Fluor 647 was then chased at 37°C for the indicated periods of time in the continuous presence of the rapalogue. Cells were analyzed as described for b. All images were captured by using the same imaging parameters. The asterisks indicate nontubularized cells expressing only GFP-Rab5a-target, where the EGF signal is significantly lower than in the adjacent tubularized cell. (Scale bar: 10 μm.)